Exploring complex discourse to index executive functioning in older adults with cognitive impairment

TL;DR

This study explores how complex discourse tasks can reveal cognitive differences in older adults with various impairments, offering insights for better assessment and rehabilitation.

Contribution

The study introduces a discourse-based method to assess executive functioning in older adults with diverse cognitive impairments.

Findings

MCI and multiple etiology groups produced fewer critical elements in discourse compared to controls.

TBI and PSCI groups showed reduced elaborations and lower semantic density in discourse.

Lower MLU scores were observed in TBI and PSCI groups, indicating shorter utterances.

Abstract

Discourse production engages multiple cognitive domains including executive functions and is known to be disrupted in individuals with cognitive impairments. Studying discourse reveals how cognitive impairments affect daily functioning by highlighting challenges in communication. We evaluated utility of a complex discourse production task in characterizing differences across older adults (n = 90) with mild cognitive impairment (MCI), traumatic brain injury (TBI), post-stroke cognitive impairment (PSCI), multiple etiology cognitive impairment, and cognitively normal older adult controls as part of the Everyday Needs Assessment for Cognitive Tasks (ENACT) study. Using a mixed-methods, cross-sectional design (n = 18 per group; mean age=71.81 ± 3.36 years) participants completed a discourse production task requiring executive processes (e.g., abstraction, planning) along with standard cognitive assessments. Discourse was analyzed using manual and automated methods. Manual coding identified critical elements (travel, activity, packing, and stay) and elaborations within categories. Automated analysis via CLAN software extracted linguistic metrics such as mean length of utterance (MLU) and semantic density. Group differences were analyzed using ANCOVA with age as a covariate. Significant group differences (p < 0.05), controlling for multiple comparisons, emerged across several discourse measures. The MCI and multiple etiology groups produced fewer critical elements compared to controls, whereas all other clinical groups showed reduced elaborations and semantic density. Additionally, participants in the TBI and PSCI groups had shorter utterances reflected in the lower MLU scores. Findings suggest that the discourse task distinguishes group-level differences and offers clinically relevant insights to guide assessment of functional challenges and inform targeted cognitive rehabilitation strategies including technology support.